Node marginal electricity price calculation method using slack variable and matrix formula
Through the LMP model based on DC optimal current and the slack variable matrix formula, the problem of difficult prediction of transmission blockage and LMP changes under load changes is solved, efficient power system planning and load management are realized, the calculation process is simplified, and the power system is improved operational efficiency.
Patent Information
- Application Number
- CN202410915201.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-09
- Publication Date
- 2024-11-22
AI Technical Summary
The existing node marginal pricing (LMP) method cannot effectively identify transmission blockage and LMP step changes when the system load changes greatly, resulting in difficult prediction and management of blockages. The calculation time of traditional models is long, making it difficult to meet the power system planning. need.
Build an LMP model based on DC Optimal Flow (DCOPF), use slack variables and matrix formulas to rewrite constraints, establish a new LMP model, determine the critical load level (CLL) and blocking lines, and calculate the power generation scheduling plan at different load levels and The marginal electricity price of the nodes, which in turn affects the transmission blockage and marginal price.
It realizes efficient identification of transmission blockage and LMP and load relationship under load conditions, simplifies the calculation process, improves the efficiency and accuracy of power system planning, and can quickly obtain the impact of different load levels on transmission blockage and marginal price.
Smart Images

Figure CN119009939A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of power system planning, and in particular relates to a node marginal electricity price calculation method using slack variables and matrix formulas. Background Art
[0002] The node marginal pricing (LMP) method has become the main method for determining node prices and managing transmission congestion in energy market operation and planning. It is always important for system operators and planners to know the future prices and the new constraints that may appear as the system load grows. This information can be used for short-term and long-term congestion relief and load management. At the same time, it is also important for generation companies to predict future electricity prices and possible congestion. The transmission is fully modeled using a market simulator based on the optimal power flow (OPF). Previous research works on the sensitivity of LMP to load are only applicable to the case where the load changes are small without tightening the constraints. It does not work when the system load changes are large, resulting in new transmission congestion and LMP step changes. Therefore, it will be very beneficial for power system planners if the transmission congestion situation or node marginal electricity price under load changes is easily available. It is relatively time-consuming to obtain the relationship curve between LMP and load using the traditional LMP model, so a more efficient algorithm is urgently needed. Summary of the invention
[0003] In view of the above problems, the present invention constructs an LMP model based on DC optimal power flow, explores an efficient algorithm suitable for short-term planning and operation of the power market to identify transmission congestion and the relationship between LMP and load level, and uses this algorithm to calculate the generation dispatch plan, blocked lines and node marginal electricity prices under the new critical load level (CLL), so as to obtain the impact of different system load levels on transmission congestion and marginal prices.
[0004] To this end, the present invention proposes a method for calculating node marginal electricity price using slack variables and matrix formulas. The method comprises the following steps:
[0005] Step A. Establish a LMP model based on DCOPF, rewrite all constraints using slack variables and matrix formulas, consider the impact of load change patterns, obtain corresponding sensitivities, and establish a new LMP model;
[0006] Step B: Based on the new LMP model, determine new tight constraints and CLL as well as new loose constraints, and determine the LMP model of a single node under the new CLL.
[0007] Preferably, the step A comprises:
[0008] Step A1: Establish a LMP model based on DCOPF, and the objective function is:
[0009]
[0010] Constraints:
[0011]
[0012] G i min ≤G i ≤G i max
[0013] Where: N is the number of nodes; C i is the power generation cost of node i; G i is the active power generated by the generator at node i; D i is the load of node i; GSF k-i is the generation shift factor from node i to line k; F k max is the transmission limit of line k;
[0014] The LMP of node i is:
[0015]
[0016] Where: M is the number of lines; LMP i energy is the price of electrical energy, LMP i cong is the blocking price, LMP i loss is the network loss price; λ is the Lagrange multiplier of the equality constraint; μ k is the Lagrange multiplier of the transport constraint; DF i is the delivery coefficient of node i;
[0017] Step A2: Apply slack variables to all inequality constraints to transform them into equality constraints; rewrite all constraints using matrix formulas so that marginal variables are expressed in terms of non-marginal variables, and obtain energy balance constraints:
[0018]
[0019] Where, MG = N MG ×1vector, the output of the marginal generator; NG=N NG ×1vector, the output of the non-marginal generator; UL=M UL ×1vector, slack variable for non-blocking circuit; CL=M CL ×1vector, represents the slack variable of the blocked line; D=[D1D2...D N ] T=N×vector, representing the matrix of the marginal unit GSF of the congested line; P = [A] - 1 p; [Q] = [A] -1 [q]; [R] = [A] -1 [r]; [T] = [A] -1 [t],
[0020]
[0021] Where: [A 11 ] represents the matrix of the marginal unit GSF of the congested line; [A 12 ]=[0] represents the 0 matrix; [A 21 ] is the matrix of the marginal unit GSF of the non-blocking line; [A 22 ]=[I] is the unit matrix; p1 represents the line flow limit of the blocked line; p2 represents the line flow limit of the non-blocked line; [q1] represents the GSF matrix of the blocked line, including all load nodes; [q2] represents the GSF matrix of all nodes of the non-blocked line; [r1] represents the negative value of the GSF of the non-marginal unit node to the blocked line; [r2] represents the GSF matrix of the non-marginal unit node to the non-blocked line; [t1]=[-I] is the negative unit matrix; [t2]=[0] is the zero matrix;
[0022] Objective function:
[0023]
[0024] Where C MG is the column vector of marginal unit power generation cost; C MG The transposed vector of NG is the column vector of power generation cost of non-marginal units; [R MG ] is the N of the R matrix MG First line; [T MG ] is the N of the T matrix MG First line,
[0025] Step A3: Consider the impact of load change pattern and obtain the sensitivity of generation dispatch; assuming that the participation factor is linear, the load can be rewritten as follows:
[0026]
[0027] ΔD=f×ΔD ∑
[0028] Where: is the load growth participation factor,
[0029] f=[f1 f2 ... f N ] T , is an N*1 column vector; ΔD is a column vector; D i is the load of node i, is the original load of node i; ΔD i The change in the load of node i; D ∑ is the total load; is the total initial load; ΔD Σ is the total amount of load change; ΔD is the column vector of load change;
[0030] Taking into account the change of system load, we can get:
[0031]
[0032] Considering the linear participation factor of the load change mode, that is, ΔD=f×AD ∑ , the above equation can be rewritten as:
[0033]
[0034] Where Q′=[Q]×f;
[0035] Further decoupling the above formula, we get:
[0036] ΔMG=Q′ MG ×ΔD ∑ +[R MG ]×ΔNG+[T MG ]×ΔCL
[0037] ΔUL=Q′ UL ×ΔD ∑ +[R UL ]×ΔNG+[T UL ]×ΔCL
[0038] From the above two equations, the sensitivity of MG and UL relative to the load can be obtained:
[0039]
[0040] The increment of the objective function:
[0041]
[0042] The sensitivity of the objective function to NG and CL can be expressed as:
[0043]
[0044] 3. A method for calculating node marginal electricity price using slack variables and matrix formulas according to claim 2, characterized in that step B comprises:
[0045] Step B1: Determine new tight constraints and new CLL;
[0046] The maximum increase in load allowed before a line reaches its limit is:
[0047]
[0048] The allowable load increase corresponding to the marginal generator is:
[0049]
[0050] By finding the minimum value from the above formula, the new CLL is equal to is the load limit allowed by the line, UL K is the slack variable of the non-blocking circuit, Q′ UL is the sensitivity coefficient of UL relative to the load, is the slack variable of the i-th marginal generator;
[0051] Step B2: Determine new loose constraints; the new loose constraints are generators or lines:
[0052] (1) Assuming that the marginal unit becomes a non-marginal unit, the load change must be offset by the previous non-marginal unit output or the previous tight slack variable, which yields:
[0053]
[0054] In the formula They are [R MG ]、[T MG ]’s,th row vector; is the sensitivity coefficient of the marginal unit relative to the load;
[0055] Expected incremental cost vector:
[0056]
[0057] In the case of load growth, the smallest positive value is selected, or in the case of load reduction, the largest negative value is selected. The corresponding j or k is the new marginal unit or the new non-blocking line.
[0058] (2) Assuming that the rth non-blocking line becomes a blocking line, we can obtain:
[0059]
[0060] Expected incremental cost vector:
[0061]
[0062] Select the smallest positive value in case of load growth or the largest negative value in case of load decrease, and the corresponding j or k is the new marginal unit or the new unblocked line;
[0063] Step B3: Determine the LMP under the new CLL using the participation factor, expressed as f = [00...1...00] T ; When the load exceeds CLL, new MG, NG, UL and CL are formulated, as well as new vectors due to changes in marginal units. The LMP at a specific node is:
[0064]
[0065] Where: is the load change on node i. BRIEF DESCRIPTION OF THE DRAWINGS
[0066] Figure 1 A flowchart of a method for calculating node marginal electricity price using slack variables and matrix formulas according to the present invention;
[0067] Figure 2 This is the electrical wiring diagram of the PJM5 node system. DETAILED DESCRIPTION
[0068] The present invention will be further described in detail below in conjunction with the accompanying drawings.
[0069] like Figure 1 As shown, it is a flowchart of a method for calculating node marginal electricity price using slack variables and matrix formula of the present invention, which includes the following steps:
[0070] Step A: Establish a LMP model based on DCOPF, rewrite all constraints using slack variables and matrix formulas, consider the impact of load change patterns, obtain corresponding sensitivities, and establish a new LMP model;
[0071] Step A1: Establish a LMP model based on DCOPF.
[0072] Objective function:
[0073]
[0074] Constraints:
[0075]
[0076] G i min ≤G i ≤Gi max (4)
[0077] Where: N is the number of nodes; C i is the power generation cost of node i; G i is the active power generated by the generator at node i; D i is the load of node i; GSF k-i is the generation shift factor from node i to line k; F k max is the transmission limit of line k.
[0078] The present invention makes the following assumptions and simplifications when establishing the model to facilitate the derivation of the formula (in practical applications, more accurate formulas can be obtained by adding constraints):
[0079] (1) Assume that each node has only one generator and one load.
[0080] (2) Generation cost / bid has a piecewise linear curve, which is consistent with industrial practice.
[0081] (3) It is assumed that there is only one cost curve (or block) for each unit; the costs of any two units should be different to avoid multiple solutions.
[0082] (4) The derivation and equations in the present invention ignore the direction of the constraint, that is, only the positive direction is considered. When dealing with practical problems, a pair of constraint equations should be constructed to deal with the positive and negative directions, and both directions should be considered at the same time.
[0083] (5) The general formula using the generation shift factor (GSF) can solve not only the transmission thermal limitation problem, but also the safety (emergency) and nomogram limitation problems.
[0084] The LMP formula for node i is:
[0085]
[0086] Where: M is the number of lines; LMP i energy is the price of electrical energy, LMP i cong is the blocking price, LMP i loss is the network loss price; λ is the Lagrange multiplier of the equality constraint, which can be obtained by taking the partial derivative of the system energy balance equation in equation (2); μ k is the Lagrange multiplier of the transport constraint; DF i is the delivery coefficient of node i.
[0087] Step A2: Apply slack variables to all inequality constraints to transform them into equality constraints. Then, it is proposed to rewrite all constraints using matrix formulas so that marginal variables are expressed in terms of non-marginal variables. Assume that the number of marginal units and the number of blocked lines satisfy the following equation:
[0088] N MG =M CL +1=MM UL +1 (6)
[0089] Where: N MG is the number of marginal units, M CL is the number of blocked lines, M UL is the number of unblocked lines.
[0090] The following energy balance equation constraints can be obtained:
[0091]
[0092] Where: MG = N MG ×1vector is the output of the marginal generator; NG=N NG ×1vector is the output of the non-marginal generator.
[0093] By introducing non-negative slack variables, the transport inequality constraints are transformed into equality constraints. k to represent the slack variables of non-blocking circuits and to use CL k To represent the slack variables of the blocked line, we can get:
[0094]
[0095] Where: CL is the set of blocked lines; UL is the slack variable of the non-blocked lines; D = [D1D2...D N ] T ; CL represents the slack variable of the blocking line, which is a zero vector in normal case.
[0096] Rewrite the above formula in matrix form as:
[0097]
[0098] Where: [A 11 ]=[GSF CL-MG ]=M CL ×N MG [A 11 ] represents the matrix of the marginal unit GSF of the congested line; [A 12 ]=[0] represents the 0 matrix; [A 21 ]=M UL ×N MGis the matrix of the marginal unit GSF of the non-blocking line; [A 22 ]=[I] is the unit matrix; p1 represents the line flow limit of the blocked line; p2 represents the line flow limit of the non-blocked line; [q1] represents the GSF matrix of the blocked line, including all nodes (load-bearing nodes); [q2] represents the GSF matrix of the non-blocked line (including all nodes); [r1] represents the negative value of the GSF of the non-marginal unit node to the blocked line; [r2] represents the GSF matrix of the non-marginal unit node to the non-blocked line; [t1]=[-I] is the negative unit matrix; [t2]=[0] is the zero matrix.
[0099] Formula (9) can be rewritten as:
[0100]
[0101] It should be noted that the slack variable vector of the current blocked line is 0 at the current load level, but it needs to be kept as a set of variables in the formula instead of a constant 0 because this situation may change as the load grows to exceed the next CLL. In other words, as the load changes, the blocked line may become a non-blocking line. This is the key step in the following analysis.
[0102] Further simplify the above formula and find the inverse of the matrix, that is:
[0103]
[0104] in is a square matrix, so we only need to calculate [A′ 11 ] is the inverse of [A′ 11 ]Since there are only a few marginal units, the size of is usually very small, which does not bring too much computational burden to the algorithm. [A]By inverting the matrix, we can get:
[0105]
[0106] In the formula
[0107] P=[A] -1 p; [Q] = [A] -1 [q]; [R] = [A] -1 [r]
[0108] [T]=[A] -1 [t]
[0109] From the above formula, we can see that formula (12) consists of three parts:
[0110] (1) Energy balance equation;
[0111] (2)McL (=N MG -1) represents the equation of the blocking line with zero slack variable;
[0112] (3)M UL Represents the equation of a non-blocking line with non-zero slack variable UL.
[0113] This formula is written in matrix form, similar to the simplex method for solving linear programming problems. The reason for this is to rewrite the non-zero variables (or basic variables) such as MG and UL to the left, and then any small change in load can be expressed as a corresponding change in MG or UL. Therefore, the objective function can be written without MG as follows:
[0114]
[0115]
[0116] Where: C MG is the marginal generation cost column vector; C MG The transposed vector of NG is the column vector of non-marginal generation cost; [R MG ] is the N of the R matrix MG First line; [T MG ] is the N of the T matrix MG First row.
[0117] Step A3: Consider the impact of load change pattern and obtain the sensitivity of power generation scheduling. The load can be rewritten as follows:
[0118]
[0119] ΔD=f×ΔD ∑ (15)
[0120] Where: is the load growth participation factor, D i is the load of node i, is the original load of node i; ΔD i The change in the load of node i; D Σ is the total load; is the total initial load; ΔD Σ is the total amount of load change; f=[f1 f2 ... f N ] T , is an N*1 column vector; ΔD is the column vector of load change.
[0121] In the above load variation model, the variation of each node load follows a linear participation factor relative to the system load variation. This model is reasonable because each node load can be modeled as its own variation pattern, so that different load characteristics such as industrial load, commercial load and residential load can be modeled accordingly. It is also flexible because the variation of each node load is independent of the initial load. This model is of great significance for short-term planning.
[0122] Taking into account the change of system load, we can get:
[0123]
[0124] Also taking into account the linear participation factor of the load change mode, that is, AD = f × ΔD ∑ , we can rewrite the above equation as:
[0125]
[0126] Where Q′=[Q]×f.
[0127] Further decoupling the above formula, we get:
[0128] ΔMG=Q′ MG ×ΔD ∑ +[R MG ]×ΔNG+[T MG ]×ΔCL
[0129] ΔUL=Q′ UL ×ΔD ∑ +[RU L ]×ΔNG+[T UL ]×ΔCL. (18) From the above two equations, we can get the sensitivity of MG and UL relative to the load:
[0130]
[0131] The incremental rewrite of the objective function is as follows:
[0132]
[0133] Therefore, the sensitivity of the objective function to NG and CL can be expressed as:
[0134]
[0135] Step B: Based on the new LMP model, determine the new tight constraints and CLL as well as the new loose constraints, and determine the LMP of a single node under the new CLL.
[0136] Step B1: Determine new tight constraints and new CLL.
[0137] When the load increases and the marginal units remain unchanged, the output of non-marginal units should remain at the minimum or maximum, and the slack variables of the blocked lines should remain zero. In other words, ΔNG = ΔCL = 0. At the same time, the marginal generators and unblocked transmission lines will also change and may gradually approach their respective limits. The first to reach its limit will be the next constraint limit. The following formula is used to analyze the above problem:
[0138]
[0139] Since all generation output and line flows at current load levels are given by the initial OPF, it is not difficult to obtain the present values of the slack variables. In fact, many optimization solvers provide these values as output.
[0140] For non-blocking transmission lines, the slack variable is UL. We can get the load growth allowed before a line reaches the limit (i.e. reaches zero) as:
[0141]
[0142] In the formula is the load limit allowed by the line. UL K is the slack variable of the non-blocking circuit. Q′ UL This formula uses the line's slack variable and sensitivity coefficient to find the line's allowable load limit.
[0143] For the generator, the slack variable is not given explicitly in the previous formula. However, it can be considered as:
[0144]
[0145] Where S MG MG is the column vector of the marginal generator output slack variables. max is the maximum output value of the marginal generator. is the maximum output value of the i-th marginal generator. MG is the slack variable of the output of the i-th marginal generator. i is the output value of the i-th marginal generator. is the maximum output value of the i-th marginal generator.
[0146] Therefore, as the load increases, the allowable load growth corresponding to the i-th marginal generator is:
[0147]
[0148] In the formula is the allowable load limit of the line. is the slack variable of the output of the i-th marginal generator. Q′ MG is the sensitivity coefficient of the marginal unit relative to the load. By finding the minimum value given by equations (23) and (25), the minimum allowable load increase can be obtained. This equation uses the slack variable and sensitivity coefficient of the marginal unit to find the allowable load limit of the marginal unit. Therefore, the new CLL is equal to
[0149] Step B2: Determine new loose constraints. When the load increases or decreases, new loose constraints will appear on the CLL. The new constraints may be generators or lines. These two cases will be discussed below.
[0150] (1) Assuming that the lth marginal unit becomes a non-marginal unit, the load change must be offset by the previous non-marginal unit output or the previous tight slack variable. From equation (17), we can get:
[0151]
[0152] In the formula They are [R MG ][T MG ] is the lth row vector of ].
[0153] It is important to note that when the load is slightly more than the new CLL, all ΔNG j and ΔCL k There should be one and only one non-zero variable in D. This is determined by the properties of linear programming, because the solution should move from one vertex to the adjacent vertex (even the vertices or boundaries of the polyhedron itself should be moved because of D ∑ Then, which ΔNG should be determined according to the change of the objective function j or ΔCL k as the next non-zero variable.
[0154] If the jth non-marginal unit becomes a marginal unit, then from (26) we can get:
[0155]
[0156] It should be noted that the above sensitivities must give a possible change that does not violate their limits. For example, for the case of load increase, if NG j has reached its maximum value, the above sensitivity should be considered only when it is negative. And, if it is in NG i minimum value, the above sensitivity should be considered only if it is positive.
[0157] If the jth blocked line is transformed into a non-blocked line, then we can get from (26):
[0158]
[0159] Similarly, only in CL k The above sensitivity is considered only when it is not pushed to negative values. k Exactly zero.
[0160] Then, from Equation 21, we can easily calculate the expected incremental cost vector, and expand it to get:
[0161]
[0162] Right now
[0163]
[0164] Finally, we can select the smallest positive value in the case of load growth (or the largest negative value in the case of load decrease), and the corresponding j (or k) is the new marginal unit (or new non-blocking line).
[0165] (2) Assume that the rth non-blocking line becomes a blocking line: Similar to the above, we can get:
[0166]
[0167] If the th non-marginal unit becomes a marginal unit, then we can get from (31):
[0168]
[0169] If the kth blocked line becomes a non-blocking line, then from (31) we can get:
[0170]
[0171] Next, the incremental cost vector can be calculated:
[0172]
[0173] Similarly, when the load increases, the smallest positive value is selected (or the largest negative value is selected when the load decreases), and the corresponding j (or k) is the new marginal unit (or the new unblocked line).
[0174] It is worth mentioning that after the introduction of a new marginal unit, the sensitivity of all existing marginal units will also change. This can be quantitatively calculated as:
[0175]
[0176] If there are no new marginal units on the CLL (e.g. load reduction, turning a blocked line into a non-blocked line), then There is only one variable is not zero; similarly, if there is a new marginal unit, then There is only one variable Not zero.
[0177] Step B3: Determine the LMP under the new CLL using the participation factor, expressed as f = [00...1...00] T , because LMP is position-dependent.
[0178] When the load exceeds CLL, we can first formulate new MG, NG, UL and CL, as well as new vectors due to marginal unit changes, etc. Then, we can use equation (17) to calculate the marginal unit relative to the load change of a single node, and the LMP at a specific node is:
[0179]
[0180] Where: is the load change on node i.
[0181] In order to enable those skilled in the art to better understand the present invention and appreciate the advantages of the present invention over the prior art, the applicant further explains the present invention in conjunction with specific embodiments.
[0182] Example
[0183] Below, a slightly modified PJM 5-Bus system will be used to illustrate the steps for determining a new tight constraint, a new loose constraint, and a new LMP as the load grows. The modifications to the original PJM 5-Bus system are as follows:
[0184] (1) The cost of the Sundance unit on Bus D was changed from $30 / MWh to $35 / MWh to differentiate it from the cost of the Solitude unit;
[0185] (2) Assume that the upper limit of line AB is 400MW.
[0186] For simplicity, we assume that the system load changes proportionally at each node load. The load change is evenly distributed across nodes B, C, and D, which all have 300MW load in the base case. Figure 2 The configuration of the system is shown, Table 1 shows the line impedance and flow limit, and Table 2 shows the GSF of AB line and ED line.
[0187] Table 1 Line impedance and flow limit
[0188]
[0189] Table 2 GSF of AB line and ED line
[0190]
[0191] The basic OPF model of economic dispatch can be written as
[0192] minC1·G1+C2·G2+C3·G3+C4·G4+C5·G5
[0193]
[0194] After solving for the OPF for the initial case (load = 900MW), there are 2 marginal and 3 non-marginal units, 5 non-blocking lines and 1 blocking line. Therefore, we have 7 non-zero basic variables, 2 for marginal units and 5 for non-blocking lines. We can then rewrite the above equation in matrix form, as shown in the equation at the bottom of the page, where MG1 and MG2 represent the Sundance and Brighton units respectively; NG1, NG2, and NG3 represent the Alta, Park City, and Solitude units respectively.
[0195]
[0196] The above equation can be rewritten as:
[0197]
[0198] At the current operating point (D ∑ =900), we get the following results from the initial OPF:
[0199]
[0200] Next, we will take load increase as an example to illustrate this process.
[0201] A. Calculate the next tight constraint
[0202] Assume a load change of ΔD ∑ , we have the following equation:
[0203]
[0204] The allowable load growth corresponding to each non-blocking line can be obtained from formula (23):
[0205]
[0206] Considering the maximum capacity, the load growth allowed for each marginal generator is:
[0207]
[0208] Similarly, for the minimum capacity, the allowable load growth of each marginal generator is:
[0209]
[0210] Therefore, the minimum positive value of load growth allowed is 63.94 MW, corresponding to the blocking of positive line flow AB. The next constraint limit will be line AB at a load of 963.94 MW.
[0211] B. Find the new marginal unit (load = 963.94MW)
[0212] When the system load increases to 963.94MW, the transmission limit of line AB is reached, and the new non-marginal generator sensitivity can be obtained from formula (27):
[0213]
[0214] Similarly, we can get:
[0215]
[0216] The incremental cost vector of NG is:
[0217]
[0218] If we compare all the incremental costs associated with NG and UL, the smallest positive value is 2.6011, which corresponds to the Solitude unit on bus C, and therefore, no new blocking line in this case.
[0219] C. Calculate the new LMP at a load level of 963.94MW
[0220] For the new marginal unit set, apply equation (18) and To calculate the load change of each bus. Again, the load change here only occurs on a specific bus.
[0221]
[0222] Similarly, we can get the LMP of all buses from 963.94MW to the next CLL
[0223]
[0224] The above embodiments describe the technical solutions of the present invention in detail. Obviously, the present invention is not limited to the described embodiments. Based on the embodiments of the present invention, people familiar with the technical field can also make various changes accordingly, but any changes that are equivalent or similar to the present invention belong to the scope of protection of the present invention.
Claims
1. A method for calculating node marginal electricity price using slack variables and matrix formula, characterized in that: The following steps are involved: Step A. Establish a LMP model based on DCOPF, rewrite all constraints using slack variables and matrix formulas, consider the impact of load change patterns, obtain corresponding sensitivities, and establish a new LMP model; Step B: Based on the new LMP model, determine new tight constraints and CLL as well as new loose constraints, and determine the LMP model of a single node under the new CLL.
2. A method for calculating node marginal electricity price using slack variables and matrix formulas according to claim 1, characterized in that: The step A comprises: Step A1: Establish a LMP model based on DCOPF, and the objective function is: Constraints: G i min ≤G i ≤G i max Where: N is the number of nodes; C i is the power generation cost of node i; G i is the active power generated by the generator at node i; D i is the load of node i; GSF k-i is the generation shift factor from node i to line k; F k max is the transmission limit of line k; The LMP of node i is: Where: M is the number of lines; is the price of electrical energy, is the blocking price, is the network loss price; λ is the Lagrange multiplier of the equality constraint; μ k is the Lagrange multiplier of the transport constraint; DF i is the delivery coefficient of node i; Step A2: Apply slack variables to all inequality constraints to transform them into equality constraints; rewrite all constraints using matrix formulas so that marginal variables are expressed in terms of non-marginal variables, and obtain energy balance constraints: Where, MG = N MG ×1vector, the output of the marginal generator; NG=N NG ×1vector, the output of the non-marginal generator; UL=M UL ×1vector, slack variable for non-blocking circuit; CL=M CL ×1vector, represents the slack variable of the blocked line; D=[D1D2...D N ] T =N×vector, representing the matrix of the marginal unit GSF of the congested line; P = [A] -1 p; [Q] = [A] -1 [q]; [R] = [A] -1 [r]; [T] = [A] -1 [t], Where: [A 11 ] represents the matrix of the marginal unit GSF of the congested line; [A 12 ]=[0] represents the 0 matrix; [A 21 ] is the matrix of the marginal unit GSF of the non-blocking line; [A 22 ]=[I] is the unit matrix; p1 represents the line flow limit of the blocked line; p2 represents the line flow limit of the non-blocked line; [q1] represents the GSF matrix of the blocked line, including all load nodes; [q2] represents the GSF matrix of all nodes of the non-blocked line; [r1] represents the negative value of the GSF of the non-marginal unit node to the blocked line; [r2] represents the GSF matrix of the non-marginal unit node to the non-blocked line; [t1]=[-I] is the negative unit matrix; [t2]=[0] is the zero matrix; Objective function: In the formula, C MG is the column vector of marginal unit power generation cost; C MG The transposed vector of NG is the column vector of power generation cost of non-marginal units; [R MG ] is the N of the R matrix MG First line; [T MG ] is the N of the T matrix MG First row, [T] = [A] -1 [t]; Step A3: Consider the impact of load change pattern and obtain the sensitivity of generation dispatch; assuming that the participation factor is linear, the load can be rewritten as follows: ΔD=f×ΔD ∑ Where: is the load growth participation factor, f=[f1 f2 … f N ] T , is an N*1 column vector; AD is a column vector; D i is the load of node i, is the original load of node i; ΔD i The change in load of node i; D Σ is the total load; is the total initial load; ΔD Σ is the total amount of load change; AD is the column vector of load change; Taking into account the change of system load, we can get: Considering the linear participation factor of the load change mode, that is, ΔD = f × ΔD ∑ , the above equation can be rewritten as: Where Q' = [Q] × f; Further decoupling the above formula, we get: ΔMG=Q′ MG ×ΔD ∑ +[R MG ]×ΔNG+[T MG ]×ΔCL ΔUL=Q′ UL ×ΔD ∑ +[R UL ]×ΔNG+[T UL ]×ΔCL From the above two equations, the sensitivity of MG and UL relative to the load can be obtained: The increment of the objective function: The sensitivity of the objective function to NG and CL can be expressed as:
3. A method for calculating node marginal electricity price using slack variables and matrix formulas as claimed in claim 2, characterized in that: The step B comprises: Step B1: Determine new tight constraints and new CLL; The maximum increase in load allowed before a line reaches its limit is: The allowable load increase corresponding to the marginal generator is: By finding the minimum value from the above formula, the new CLL is equal to is the load limit allowed by the line, UL K is the slack variable for the non-blocking circuit, Q' UL is the sensitivity coefficient of UL relative to the load, is the slack variable of the i-th marginal generator; Step B2: Determine new loose constraints; the new loose constraints are generators or lines: (1) Assuming that the lth marginal unit becomes a non-marginal unit, the load change must be offset by the previous non-marginal unit output or the previous tight slack variable, which yields: In the formula They are [R MG ]、[T MG ]’s l-th row vector; is the sensitivity coefficient of the marginal unit relative to the load; Expected incremental cost vector: In the case of load growth, the smallest positive value is selected, or in the case of load reduction, the largest negative value is selected. The corresponding j or k is the new marginal unit or the new non-blocking line. (2) Assuming that the rth non-blocking line becomes a blocking line, we can obtain: Expected incremental cost vector: Select the smallest positive value in case of load growth or the largest negative value in case of load decrease, and the corresponding j or k is the new marginal unit or the new unblocked line; Step B3: Determine the LMP under the new CLL using the participation factor, expressed as f = [00...1...00] T ; When the load exceeds CLL, new MG, NG, UL and CL are formulated, as well as new vectors due to changes in marginal units. The LMP at a specific node is: Where: is the load change on node i.
